Effort: max · Max
zai-orgReleased Aug 18, 2026
GLM-5.3
Reuses the GLM-5.2 base with new post-training; exposes low, high and max reasoning effort, defaulting to max.
Source description
GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves on GLM-5.2 in coding and in the balance between performance and token efficiency.
Context window1M
Max output943.7K
Provider endpoints59
Benchmark results19
Agent runs1
Measured performance
All leaderboards ↗Benchmark results
Effort: max · Max
GDPval-AA ↗2.1
1,650.31Effort: max · Max
AutomationBench-AA ↗AA Index 4.3.2 collection
62.2%Effort: max · Max
Effort: max · Max · Harness unreported
SciCode ↗AA Index 4.3.2 collection
59.03%AA · Under reviewEffort: max · Max
Humanity’s Last Exam ↗AA Index 4.3.2 collection
42.26%Effort: max · Max
GDP.pdf ↗AA Index 4.3.2 collection
11.2%Effort: max · Max
CritPt ↗AA Index 4.3.2 collection
19.14%AA · Under reviewEffort: max · Max
AA-Omniscience ↗AA Index 4.3.2 collection
14.3Effort: max · Max
AA-LCR ↗1.1
79.67%Effort: max · Max
LiveBench Overall ↗2.0 · 2026-10-14 (dataset 2026-09-02)
60.74%Effort: max · single run · equal mean of 7 suites
LiveBench Design ↗2.0 · 2026-10-14 (dataset 2026-09-02)
77.3%Effort: max · single run
LiveBench Research ↗2.0 · 2026-10-14 (dataset 2026-09-02)
73.2%Effort: max · single run
LiveBench Browser Use ↗2.0 · 2026-10-14 (dataset 2026-09-02)
0%Effort: max · single run
LiveBench Coding ↗2.0 · 2026-10-14 (dataset 2026-09-02)
82.8%Effort: max · single run
LiveBench Agentic ↗2.0 · 2026-10-14 (dataset 2026-09-02)
35.9%Effort: max · single run
LiveBench Data Analysis ↗2.0 · 2026-10-14 (dataset 2026-09-02)
70.2%Effort: max · single run
LiveBench Reasoning ↗2.0 · 2026-10-14 (dataset 2026-09-02)
85.8%Effort: max · single run
Provider measurements
1 configurationSpeed
≈10K input tokens1 requestmedian / 72h
Artificial Analysis · first-party providers · serving endpoint not reported
Z AI
GLM-5.3 (Max)
Reasoning max
- Output rate
- 80.6004 tok/s
- Time to first chunk
- 2.72669 s
Last checked Source ↗
Methodology 2.2.0 & measurement details
First chunk includes request processing and network time, and may include reasoning.
- Prompt
- long · ≈10,000 input tokens
- Parallel requests
- 1
- Minimum answer
- 1,500 tokens
- Aggregation
- median over 72 hours
- Token counting
- o200k_base
- Benchmark client
- Google Cloud us-central1-a
- Serving endpoint
- Not reported
- Serving region
- Not reported
- Sample size
- Not reported
- Window end
- Not reported
Model + harness
All agent results ↗Agent runs
Terminal-Bench 4.0
claude-code2.1.207
Reasoning effort max
Tested model identity
GLM-5.3anthropic/glm-5.3Pass rate41.82%
95% CI 38.59%–45.05%
n = 330Checked
Access & cost
60 quotesPrices across providers
Your token mix
USD · listed token ratesOpenRouter
Catalog quote- Input
- $0.079
- Output
- $4.80
- Cached input
- $0.078
novita
via Hugging Face- Input
- $1.40
- Output
- $4.40
together
via Hugging Face- Input
- $1.40
- Output
- $4.40
fireworks-ai
via Hugging Face- Input
- —
- Output
- —
featherless-ai
via Hugging Face- Input
- —
- Output
- —
zai-org
via Hugging Face- Input
- —
- Output
- —
baseten
via Hugging Face- Input
- $1.40
- Output
- $4.40
deepinfra
via Hugging Face- Input
- $0.90
- Output
- $4.00
- Input
- $0.90
- Output
- $4.00
Experiential
Standard list rate
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
Catalog list rate. Plan allowances and temporary discounts are separate.
Together AI
Serverless · quantization=FP4
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
Organization: Z.ai
Fireworks AI
Standard · service_tier=default
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
Fireworks AI
Standard · service_tier=priority
- Input
- $1.75
- Output
- $5.50
- Cached input
- $0.325
Fireworks AI
Fast · service_tier=default
- Input
- $2.10
- Output
- $6.60
- Cached input
- $0.39
- Input
- $0.0632
- Output
- $4.40
- Cached input
- $0.0588
- Input
- $0.079
- Output
- $4.80
- Cached input
- $0.078
- Input
- $0.08
- Output
- $4.381
- Cached input
- $0.076
- Input
- $0.084
- Output
- $6.00
- Cached input
- $0.052
- Input
- $0.14
- Output
- $4.20
- Cached input
- $0.105
Source discount 30%; rates as published
- Input
- $0.14
- Output
- $4.40
- Cached input
- $0.10
- Input
- $0.17
- Output
- $4.80
- Cached input
- $0.169
- Input
- $0.18
- Output
- $2.95
- Cached input
- $0.15
Endpoint degraded (status -5)
- Input
- $0.19
- Output
- $4.40
- Cached input
- $0.19
- Input
- $0.49
- Output
- $2.60
- Cached input
- $0.19
- Input
- $0.50
- Output
- $1.505
- Cached input
- $0.119
- Input
- $0.50
- Output
- $1.505
- Cached input
- $0.119
- Input
- $0.55
- Output
- $1.505
- Cached input
- $0.18
- Input
- $0.5625
- Output
- $2.50
- Cached input
- $0.125
Source discount 37.5%; rates as published
- Input
- $0.5992
- Output
- $1.8832
- Cached input
- $0.1113
Source discount 57.2%; rates as published
- Input
- $0.60
- Output
- $2.50
- Cached input
- $0.10
- Input
- $0.70
- Output
- $2.20
- Cached input
- $0.13
Source discount 50%; rates as published
- Input
- $0.70
- Output
- $2.20
- Cached input
- $0.13
Source discount 50%; rates as published
- Input
- $0.84
- Output
- $2.64
- Cached input
- $0.156
Source discount 40%; rates as published
- Input
- $0.84
- Output
- $2.66
- Cached input
- $0.19
- Input
- $0.91
- Output
- $2.86
- Cached input
- $0.169
- Input
- $0.98
- Output
- $3.08
- Cached input
- $0.182
Source discount 30%; rates as published
- Input
- $1.19
- Output
- $3.74
- Cached input
- $0.238
- Input
- $1.20
- Output
- $4.00
- Cached input
- $0.19
- Input
- $1.26
- Output
- $3.96
- Cached input
- $0.234
Source discount 10%; rates as published
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.14
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.14
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.14
- Input
- $1.40
- Output
- $4.40
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.13
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
Source discount 20%; rates as published
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.14
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.40
- Output
- $4.40
- Cached input
- $0.26
- Input
- $1.54
- Output
- $4.84
- Cached input
- $0.154
- Input
- $2.10
- Output
- $6.60
- Cached input
- $0.40
- Input
- $2.10
- Output
- $6.60
- Cached input
- $0.39
- Input
- $2.10
- Output
- $6.60
- Cached input
- $0.21
- Input
- $2.10
- Output
- $6.60
- Cached input
- $0.21
- Input
- $2.80
- Output
- $8.80
- Cached input
- $0.56
Offers for this model
1 offerExperiential Labs · Discount
7.5% off · Experiential Pro model discounts
Paid Pro plan required, covering the first 2,000 credits of usage each month, then list price.
Checked Source ↗
Model specifications
- Developer
- zai-org
- Modality
- Language
- Context
- 1,048,576 tokens
- Maximum output
- 943,718 tokens
- Weights
- Open weights
- Released
- Aug 18, 2026
- Provider listed
- Aug 18, 2026
- First discovered
- Oct 16, 2026
Exact identities
- openrouter
z-ai/glm-5.3 - huggingface
zai-org/GLM-5.3 - deepinfra
zai-org/GLM-5.3 - experiential
glm-5.3 - artificial-analysis
cd684ea4-b475-4269-b001-d469d06d8a7a - livebench
glm-5.3
Sources
Updated Oct 17, 2026
More from zai-org
View all ↗zai-orgReleased Aug 26
GLM-5.3-Flash
GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context…
1M contextImage inputVideo input
No reported score
Price not listed
No reported score
Price not listed
zai-orgListed Jun 16
GLM-5.2
GLM-5.2 is Z-AI's latest flagship model for long-horizon tasks. It marks a substantial leap in long-horizon task capability over its predecessor GLM-5.1 and, for the first time, delivers that capability on…
1M contextTool callingStructured outputNo reported score
No reported score